Eating Compulsivity in Inpatients with Severe Obesity and the General Population: The Italian Version of the Measure of Eating Compulsivity (MEC10-IT)

This study aimed to validate and investigate the psychometric properties of the Italian version of the Measure of Eating Compulsivity-10 (MEC10-IT) in a sample of inpatients with severe obesity (Study 1), and to test the measurement invariance of the measure across non-clinical and clinical samples (Study 2). In the first study, a confirmatory factorial analysis (CFA) was conducted among 452 patients in order to confirm the factorial structure of the MEC10-IT. In the second study, the psychometric properties of the MEC10-IT were tested on 453 inpatients with severe obesity and a community sample of 311 participants. The CFA confirmed the factorial structure of the MEC10-IT among an Italian sample of adult inpatients with severe obesity (Study 1). The MEC10-IT was also demonstrated to be invariant between the clinical and the community sample and to possess good psychometric properties, as well as excellent screening abilities for classifying individuals with problematic eating behaviors (Study 2). In conclusion, the MEC10-IT could be considered as a valid and reliable tool for the assessment of compulsive eating in both non-clinical and clinical samples and represents a psychometrically sound measure for clinical and research purposes.


Introduction
Obesity is one of the greatest public health challenges of the 21st century, causing various physical disabilities (e.g., cardiovascular diseases, cancer, diabetes, and musculoskeletal disorders) and psychological problems [1,2]. Worldwide obesity has nearly tripled since 1975 [3], partially due to the increased availability of hyper-palatable and highly processed foods (which are high in fat and sugars with chemical flavors) [4][5][6][7]. Moreover, highly processed foods share the typical characteristics and benchmarks of addiction-target substances-such as tobacco [8].
As a result, individuals may become addicted to highly processed foods and thus may be driven to engage in dysfunctional behaviors characterized by an irresistible, compulsive urge to eat highly processed food. More in detail, behavior compulsivity-a persistent drive to enact a behavior (e.g., drug use and food consumption) and the inability to control it-is a central feature of both Substance-Related and Addictive Disorders (SRAD) and behavioral addictions [19,20], as well as of certain eating disorders (ED)-e.g., binge eating disorder (BED) and bulimia nervosa (BN). Indeed, certain food-related psychopathologies, such as FA, BED, and BN, are centrally associated with an irresistible, uncontrollable urge to overeat, despite efforts to control this behavior [21,22].
Furthermore, research shows how compulsive (over-) eating behaviors may be dissected into three main facets [20], each of which has analogies with specific aspects of EDs and/or SRADs. The first facet is habitual overeating. Repeated experiences of eating palatable food can establish ordered and structured action sequences so that certain contexts or stimuli that are associated with food easily evocate compulsive overeating. In other words, environmental stimuli associated with food (e.g., the smell of bread, supermarkets, and food commercials) can augment the intense desire for a certain food, as well as "food-seeking" behaviors, regardless of real hunger [23][24][25]. The second facet is related to negative affect. Compulsive eating behaviors often aim to prevent or relieve emotional distress, such as anxiety and sadness, thus triggering a vicious circle between eating and negative affect [26][27][28][29][30]. The third facet is excessive eating despite the awareness that it can lead to several medical conditions that are associated with excessive weight (e.g., cardiovascular disease, type II diabetes, and cancer), as well as psychological issues, psychiatric disorders, and functioning impairment [31,32]. The persistence of compulsive eating behaviors despite their negative consequences is a key point of addictive behaviors with a feeling of a "loss of control", overeating, and drug taking [33][34][35][36][37][38].
Since it was conceptualized as a related, but distinct, construct from that of addiction, much attention has been recently dedicated to defining what constitutes compulsive eating behavior, and how it should be specifically measured and treated [20]. The Brief Measure of Eating Compulsivity (MEC10) is a 10-item instrument that is specifically focused on the evaluation of compulsive eating within the FA framework. Validation of the tool showed excellent test-retest reliability and high internal consistency. Furthermore, scores on the MEC10 were found to be statistically predictive of FA diagnosis, based on SRAD criteria [39].
Since the MEC10 is currently not available for use among the Italian population, the purpose of this two-step study was to examine its psychometric properties. Specifically, a confirmatory factorial analysis (CFA) was employed to investigate the construct validity of the Italian version of the MEC10 (MEC10-IT) in a sample of inpatients with severe obesity (Study 1). Then, the measurement invariance (MI) and other psychometric proprieties of the questionnaire were tested across clinical and non-clinical samples (Study 2). This would allow the scale to be applied to multiple groups of individuals who are (potentially) prone to dysfunctional eating behaviors (such as compulsive eating).

Translation and Cultural Adaptation
According to international guidelines [40,41], the MEC10-IT was first translated from English to Italian by two bilingual translators, independently. A summary version of the tool was then obtained from the independent translations, and back-translated into English by another independent translator whose mother tongue was English to ensure equivalence between the translations. Following this, a sample of 30 participants (15 participants who met the inclusion criteria of the study and 15 participants who were retrieved from the general population) was asked to fill in the pre-final version of the MEC10-IT in order to assess its items' comprehensibility [42,43]. At this stage, no further adjustment was made (Table 1 and Appendix A). Note: SD = standard deviation; SK = skewness; K = kurtosis; λ = factor loading; R 2 = explained variance.

Sample Size Determination
Considering the aim of the present study, the sample size was set a priori according to the "n:q criterion"-where n is the number of participants and q is the number of (free) model parameters to be estimated [44][45][46]. A ratio of five participants for each parameter was, therefore, guaranteed: N minimum = 250.

Procedure
A cross-sectional research method was employed to test the factor structure of the MEC10-IT.
A sample of inpatients with severe obesity who were recruited at the IRCCS Istituto Auxologico Italiano, Ospedale San Giuseppe, Verbania, Italy, within their first week of a one-month rehabilitation program for weight reduction was asked to fill in a survey comprising selected socio-demographic questions, as well as the Italian version of the MEC10-IT, of the modified Yale Food Addiction Scale 2.0 (mYFAS 2.0), and of the Binge Eating Scale (BES).
The inclusion criteria were as follows: (A) having Italian as their first language; (B) being 18 years old or older; (C) having a BMI ≥ 35 kg/m 2 ; and (D) providing signed, written informed consent to participate in the study. The exclusion criteria were as follows: (A) the incapability to complete the survey due to vision and/or cognitive problems and (B) failure to provide answers to all items. All participants signed the written and informed consent.

Participants
A total of 474 participants were initially recruited-but 22 inpatients were left out due to missing answers.
The final sample comprised 452 inpatients with severe obesity: 206 males (45.6%) and 246 females (54.4%), aged 18 to 82 years (mean = 54.18, SD = 12.657), with a BMI ranging from 35.08 to 86.18 kg/m 2 (mean = 43.22, SD = 6.77). The MEC10-IT is a self-report questionnaire composed of 10 items answered on a 5-point Likert-type response scale (from 0 = "Very Untrue" to 4 = "Very True") assessing compulsive eating behaviors. High scores correspond to a high degree of eating compulsivity. In the first validation study [39], the MEC10 showed acceptable internal consistency, with Cronbach's alpha being equal to 0.946. The mYFAS 2.0 [47,48] is a 13-item self-report questionnaire scored on an 8-point Likert-type scale (ranging from 0 = "never" to 7 = "every day") assessing the presence of FA. Similarly to its original extended version (YFAS 2.0) [49][50][51], the mYFAS 2.0 evaluates the 11 DSM-5 diagnostic criteria for SRADs [28] and the impairment and/or distress related to food experience by the subjects over the previous 12 months-for example, (A) consumed more food than intended; (B) unable to cut down or stop; (C) use despite physical/emotional consequences; and (D) craving. The mYFAS2.0 offers the following two scoring options: the symptom count score (namely, the number of FA symptoms experienced in the previous year) and the diagnostic score (namely, the absence/presence of FA). In the present sample, the mYFAS2.0 showed satisfactory internal consistency, as the KR20 coefficient was 0.831.

The Binge Eating Scale (BES)
The BES [52,53] is a self-report measure of binge eating severity in both general [54] and clinical populations [55]. It consists of 16 items related to 2 different dimensions (FC-feelings/cognitions; and B-behaviors), and a total score [55]. The BES has received general support as a reliable and valid measure of eating-related pathology-especially for BED. It commonly shows a satisfactory internal consistency, and several studies highlight its ability to discriminate between clinical and non-clinical individuals [52]. A total score equal to, or higher than, 16 (total BES ≥ 16) is considered a clinical cutoff for BED [52,53]. In the present sample, the BES showed satisfactory internal consistency, as Cronbach's alphas were 0.890, 0.809, and 0.816 for the BES total score, the FC subscale, and the B dimension, respectively.
As reported in Table 1, all of the items' loadings were statistically significant a ranged from 0.766 (item#9) to 0.869 (item#4), with a mean that was equal to 0.824 and SD equal to 0.033.
As reported in Table 1, all of the items' loadings were statistically significant and ranged from 0.766 (item#9) to 0.869 (item#4), with a mean that was equal to 0.824 and an SD equal to 0.033.

Sample Size Determination
As in Study 1, the "n:q criterion" was used to calculate the minimum sample size [44][45][46], and a ratio of five participants for each parameter (5:1; nminimum = 250) was ensured.

Procedure
The inpatients with severe obesity (BMI ≥ 35 kg/m 2 ) were selected for inclusion in this study based on the same procedures and criteria as those used in Study 1; however, this sample differs from that of Study 1. In addition, in line with previous investigations [50,79], a community sample was recruited in Padua, Italy, through personal invitations and advertisements. A snowball sampling technique was used.
The inclusion criteria were as follows: (A) being a native Italian speaker; (B) being 18 years old or older; and (C) providing signed, written informed consent to participate in the study. The participants were, instead, excluded if they presented with vision and cognitive impairments preventing them from filling in the questionnaires.

Participants
The final sample of this study was composed of 764 participants, including 453 inpatients with severe obesity and a community sample of 311 participants. The clinical population comprised 196 males (43.3%) and 257 females (56.7%), in the age range from 18 to

Sample Size Determination
As in Study 1, the "n:q criterion" was used to calculate the minimum sample size [44][45][46], and a ratio of five participants for each parameter (5:1; n minimum = 250) was ensured.

Procedure
The inpatients with severe obesity (BMI ≥ 35 kg/m 2 ) were selected for inclusion in this study based on the same procedures and criteria as those used in Study 1; however, this sample differs from that of Study 1. In addition, in line with previous investigations [50,79], a community sample was recruited in Padua, Italy, through personal invitations and advertisements. A snowball sampling technique was used.
The inclusion criteria were as follows: (A) being a native Italian speaker; (B) being 18 years old or older; and (C) providing signed, written informed consent to participate in the study. The participants were, instead, excluded if they presented with vision and cognitive impairments preventing them from filling in the questionnaires.

Measures
The same survey that was used in Study 1 was administered to each participant, in addition to the Dutch Eating Behavioral Questionnaire (DEBQ).
The DEBQ [80,81] is a 33-item self-report measure of behaviors and attitudes related to eating disorders that is commonly used in both non-clinical [54,82] and clinical samples [83]. The responses are scored on a five-point Likert-type scale (ranging from 1 = "never" to 5 = "very often") and loaded onto the following three dimensions: emotional eating (EE), restrained eating (RE), and external eating (ExE), in addition to providing a total score. In the present sample, the Cronbach's alphas were 0.921, 0.883, 0.968, and 0.867 for the total score, the RE subscale, the EE subscale, and the ExE subscale, respectively.
In the present sample, the MEC10-IT showed adquate internal consistency: Cronbach's alphas was 0.944. Also, in this second study the mYFAS2.0 showed adquate internal consistency: Cronbach's alphas was 0.917. Lastly, also the BES showed adquate internal consistency, as Cronbach's alphas were 0.894, 0.807, and 0.823 for the BES total score, the FC subscale, and the B dimension, respectively.
A first-order single-factor model ( Figure 1) was specified for both the non-clinical and the clinical samples. The DWLS estimator was run [68,71]. The model fit was calculated with the above-mentioned model fit indices-the χ 2 , the RMSEA, the CFI, and the SRMR [68,73]. The following cutoff criteria for the goodness of fit were used: p(χ 2 ) > 0.050 ns; RMSEA ≤ 0.08; CFI ≥ 0.95; and SRMR ≤ 0.08 [68][69][70]73]. The model structure was tested on each sample independently.
Moreover, the item discriminant power (IDP) was computed in order to evaluate the ability of the items to discriminate between participants with low or high EC [94,95]. In more detail, the maximum total score and the quartile rank for each participant were calculated. Then, independent sample t-tests-and their effect size (Cohen's d) [78]-were computed to assess the IDP using the total score of the scale as the dependent variable and its lowest and highest quartile as the grouping variable [94,95]. In addition, the item-total correlation (adjusted) was computed [77,96,97].
The Pearson correlation coefficient was used to assess the convergent validity and was interpreted using Cohen's benchmarks [77,78].
The incremental validity was also examined. A first regression analysis (generalized linear model with zero-inflated negative binomial distribution) was conducted to assess the increase in the explained variance (∆pseudo-R 2 ) in the FA symptoms that were related to the MEC10-IT. A second multiple regression analysis (general linear model with continuous dependent variable) was conducted to assess the increase in the explained variance (∆R 2 ) in the binge eating tendencies that were related to the MEC10-IT. For each regression analysis, eating attitude measures were first entered into the regression equation (first block). The MEC10-IT total score was then added (second block) and the ∆R 2 was checked in order to evaluate its contribution.
Moreover, receiver operating characteristics (ROC) curves were used to assess the accurateness of the MEC10-IT to distinguish (A) the participants presenting FA from those without FA; and (B) the participants with BED from those not presenting binge eating symptoms [98,99]. In accordance with previous studies [81], an overall sample of inpatients with severe obesity was used by merging the inpatients who were enrolled in Study 1 and the inpatients who were enrolled in Study 2. Thus, the final sample that was used to run the ROC curve analysis was equal to 909 participants. The overall accuracy-validity of the MEC10-IT was assessed with the area under the ROC curve (AUC; 5000 stratified bootstrap resamples). Swets' benchmarks were used to interpret the AUC [100,101]. Furthermore, the accuracy (ACC), sensitivity (SE), and specificity (SP) were calculated for each MEC cutoff point [98,99].

Structural Validity
When combining the results of the two samples, the MEC10-IT presented a good fit to the data. The chi-square statistic was statistically significant: χ 2 (35) = 176.414; and p < 0.001. Additionally, all of the other fit indices revealed a good fit to the data: the RMSEA = 0.073; 90%CI 0.062-0.084; p(RMSEA < 0.05) < 0.001; the CFI = 0.998; and the SRMR = 0.038.

Psychometrics Properties
The IDP analysis showed that the 10 items of the MEC10-IT were able to discriminate among the participants with low and high levels of eating compulsivity in both of the samples ( Table 2). As for the sample of inpatients with severe obesity, the discrimination parameter t i ranged from |21.68| (item#1) to |33.66| (item#6), with an associated effect size (Cohen's d) ranging from |2.92| to |4.49|, respectively. In the community sample, the discrimination parameter t i ranged from |14.13| (item#10) to |25.01| (item#6), with an associated effect size (Cohen's d) ranging from |2.24| to |3.98|, respectively. In addition, the item-total correlation (adjusted) revealed a moderate-to-strong association between each item of both of the samples and the MEC10-IT total score ( Table 2).
The reliability analysis revealed satisfying results. In the sample of inpatients with severe obesity, Cronbach's alpha was 0.943, and McDonald's ω was 0.954, while among the general population, Cronbach's alpha was 0.935, and McDonald's ω was 0.953.

Incremental Validity
The first multiple regression (zero-inflated negative binomial generalized linear model; ZINB GML) was performed in order to examine the contribution of the MEC10-IT to the explained variance of the mYFAS 2.0 symptom count (Table 3). Based on the results of the correlation analysis (Figure 3), in the first block, BES FC, BES B, DEBQ ExE, and DEBQ EE entered the regression model and accounted for 47.6% (Cragg and Uhler's pseudoR 2 = 0.476) of the explained variance of the mYFAS2.0 symptom count. In the second block, the MEC10-IT entered the regression equation and was found to significantly increase the proportion of the FA symptom count explained variance that was accounted by the model: 54.9% (Cragg and Uhler's pseudoR 2 = 0.549); ∆pseudoR 2 = 0.074.
The second multiple regression (linear model, LM) was performed in order to examine the contribution of the MEC10-IT to the explained variance in the BES total score (Table 3). Based on the results of the correlation analysis (Figure 3), in the first block, the mYFAS2.0 symptom count, DEBQ ExE, and DEBQ EE entered the regression and accounted for 65.3% (adjusted R 2 = 0.653) of the BES total score. In the second block, the MEC10-IT entered the regression equation and was found to significantly increase the proportion of the explained variance of the binge eating tendencies that were accounted by the model: 72.2% (adjusted R 2 = 0.722); ∆R adj 2 = 0.069.  the mYFAS2.0 cutoffs, 657 (72.6%) of the participants were categorized as non-food-addicted, while 248 (27.4%) of the participants were considered to be food-addicted (total sample = 905). Therefore, when using a cutoff of 17 for the MEC10-IT, the ROC curves revealed that 450 (49.7%) of the participants were properly categorized as 'true negative' and 201 (22.2%) as 'true positive' (71.9%). In contrast, 47 (5.2%) of the respondents turned out to be 'false negative' and 207 (22.9%) of the participants turned out to be 'false positive' (28.1% misclassified). Based on the BES, 647 (71.5%) of the participants were classified as non-binge-eaters, while 258 (28.5%) of the participants were classified as binge eaters (total sample = 905). Therefore, when using a cutoff of 20 for the MEC10-IT, the ROC curves displayed that 543 (60%) Based on the BES, 647 (71.5%) of the participants were classified as non-binge-eaters, while 258 (28.5%) of the participants were classified as binge eaters (total sample = 905). Therefore, when using a cutoff of 20 for the MEC10-IT, the ROC curves displayed that 543 (60%) of the participants were properly categorized as 'true negative' and 216 (23.9%) as 'true positive' (83.87%). In contrast, 42 (4.6%) of the respondents turned out to be 'false negative' and 104 (11.5%) of the participants turned out to be 'false positive' (16.13% misclassified).

Discussion
Some individuals may show patterns of overeating, which exist on a continuum of severity ranging from casual indulgence to compulsive drive to consume certain foods [102] showing "addictive" tendencies toward highly processed foods similar to SRAD. In order to further understand the emerging concept of FA-and particularly the role played by its documented key component of compulsivity in the development and maintenance of obesity problems-the Brief Measure of Eating Compulsivity (MEC10) has been recently developed and tested on a sample of individuals with severe obesity in New Zealand, showing excellent psychometric proprieties [39].
Since this tool is not available for use in Italy, this study aimed to test the factorial structure of the Italian version of the MEC10 in a sample of Italian adult inpatients with severe obesity. In addition, this contribution had the purpose of offering an exhaustive examination of the psychometric properties of the questionnaire, which includes its measurement equivalence between clinical and non-clinical samples. Indeed, the comparability of the patient-reported outcome measures over different populations is essential to support clinical diagnostics, research on the quality of healthcare, and population health monitoring.
The results from the first study showed that, in a sample of adult inpatients with severe obesity, the single-factor structure of the MEC10-IT was satisfactory. In fact, the CFA showed excellent structural validity of the tool, and the reliability analysis was satisfying. Significant correlations were also found between the MEC10-IT total scores and both the mYFAS2.0 symptom count and the BES subscales, but not between the MEC10-IT and the individuals' BMI. These findings support the association between the construct of FA and compulsive eating patterns as measured by the BES and are in line with those of previous studies that postulated the absence of a linear relationship between FA and BMI [50,103].
In Study 2, the structural validity and the reliability of the questionnaire were further confirmed in both clinical and non-clinical samples, and the MEC10-IT was shown to be a reliable tool in discriminating among the participants with low and high eating compulsivity across the populations. These results suggest the ability of the items to report inter-individual differences in the individuals' compulsive behavior, as well as the ability of each item to represent its latent construct.
In the overall sample, large, significant positive correlations were also found between the MEC10-IT and both the mYFAS2.0 symptom count and the BED dimensions. In addition, the MEC10-IT was associated with the DEBQ total score, the DEBQ EE, the DEBQ ExE dimensions, and the individuals' BMI, but not with the DEBQ RE. These findings suggest FA to be an interesting theoretical framework to examine disordered eating behaviors and highlight the important role of compulsive eating in the etiopathogenesis and maintenance of overweight and obesity and eating disturbances. Still, the constructs of FA and compulsivity that are related to BMI need to be clarified in future studies.
Because testing the invariance of factorial structures between different populations is central to scale validation, MI analysis was performed to explore this difference between the inpatients with severe obesity and the general population. The results show that the 10 items comprising the MEC10-IT were equally associated with the latent factor in each sample and that both of the populations had the equivalent expected item response at the identical absolute level of the trait. These findings suggest that the respondents in both of the samples had the same understanding of the MEC10-IT items (the factor structure was equivalent), with an equivalent strength (items were equally related to the latent construct), and with an equal baseline (item thresholds were equivalent). Still, the latent trait did not have the same distribution across the samples (latent means were different). This means that the MEC10-IT can be used for clinical and research purposes in both a population of individuals with obesity and a community sample (equal items threshold); however, the results should be interpreted with caution (different latent means) [104][105][106][107][108].
In addition, the results from the ROC analyses showed that the MEC10-IT represented a good screening/diagnostic measure for the detection of FA in individuals with severe obesity. In fact, it exhibited high accuracy (AUC = 0.819), sensitivity (0.810), and specificity (0.685) in distinguishing the inpatients with severe obesity presenting with FA from those that did not show FA symptoms. Equally, the MEC10-IT was shown to be reliable in identifying the presence of binge eating in people with severe obesity, as it was highly accurate (AUC = 0.905), sensible (0.837), and specific (0.839) in distinguishing between the inpatients with severe obesity with BED and those without BED. Furthermore, it has been shown that the MEC10-IT score accounted for unique variance in the mYFAS 2.0 symptom count dimension, as well as in BED tendencies.
Still, this study presents some limitations. First, the use of a cross-sectional research design did not permit the assessment of the probable changes in the MEC10-IT scores over time or its temporal stability. Future research should fill this gap by assessing the additional psychometric properties of the tool, including longitudinal MI and test-retest reliability. Moreover, the age of the participants largely varied in both of the samples. This may affect the individuals' metabolism and compulsive eating behaviors. Future research should test possible explanatory models of compulsive eating behaviors by considering possible intervening variables (e.g., age, gender, BMI, etc.).
Regardless of the above-mentioned limitations, this contribution is the first aimed at investigating the psychometric proprieties of the MEC10-IT in both the general population and in individuals with severe obesity in Italy, confirming the reliability and the validity of the tools. The results of both of these studies are based, indeed, on solid and worldwide recommended statistics.
The MEC10-IT can, therefore, be used for clinical and research purposes to identify the presence of compulsive eating with accuracy and parsimony, thus representing a viable alternative to longer questionnaires, including the BES. Indeed, in the field of feeding disorders and EDs, eating compulsivity represents a crucial construct to be considered for more precise psychological interventions, as (1) it provides important information for both the conceptualization and the treatment of disordered eating patterns and (2) it may represent a transdiagnostic maintenance factor of EDs.
Moreover, compulsive eating behavior is a core element of either BED or FA diagnosis, but it is not required for obesity detection [109]. Still, 40-70% of individuals with BED and/or FA are likely to be overweight/obese [11,[110][111][112]. Similarly, in adults with obesity and BED, an additional diagnosis of FA is associated with greater pathology and comorbid psychiatric disorders [113]. The high degree of comorbidity among these feeding and eating disorders supports their shared etiologies and/or underlying mechanisms, manifesting through compulsive eating behaviors. In addition, research suggests a partial overlap between FA, binge eating, and eating compulsivity-signifying that these three psychological constructs may be rather interconnecting [112]. Indeed, FA may lead to compulsive (over-) eating behaviors, and these-in turn-may become chronic to the point of determining markedly dysfunctional eating habits, such as binge eating behaviors-characterized by a loss of control and a reduced quality of life [112].
This study showed that the MEC10-IT might be a valid and reliable tool for the detection and measure of compulsive (over-) eating behaviors in individuals with severe obesity, as well as in the general population. Still, given that compulsive eating is present across weight classes, future studies should investigate if these findings are generalizable to a wider range of BMIs.

Conclusions
The MEC10-IT represents a reliable tool to assess the presence (and the level) of compulsive eating patterns in both clinical and non-clinical samples. In fact, the MEC10-IT showed good construct validity and reliability in both patients with severe obesity and in the general population. Still, compulsive eating behavior is characteristic of several eating-related conditions, including BED, obesity, and FA. This self-report questionnaire can therefore be used by clinicians and researchers to promptly assess the problem of the loss of control over the food that leads to obesity and eating-related pathologies above and beyond the already existing measures of FA and binge eating [114], thus informing the clinical prediction of compulsivity-related symptoms and supporting the development of interventions that are specifically aimed at addressing compulsive eating in obesity, BED and FA. Further research should seek to replicate these findings by employing cross-cultural and longitudinal designs and examine the MEC10-IT relationship of eating compulsivity with anthropometric and metabolic characteristics, as well as across psychopathological profiles.